The evaluation of land-use/land-cover changes (LUCC) in the arid region was scarcely quantified before the 1980s due to the unavailability of satellite data, resulting in our little knowledge of the effects of LUCC on climate change and carbon/water cycle in that period. In this study, we investigated the evolution of oases in Sangong river basin since the 1950s by using seven land cover maps derived from black-and-white aerial photographs (1958, 1968 and 1978), color-infrared aerial photograph (1987), Landsat TM imagery (1998), SPOT (2004) and Landsat OLI images (2014). Results show that: (1) Since 1950, the new oasis consecutively expanded more than four times towards the bottom of the alluvial plain, causing a huge shrinking of the desert landscapes dominated by haloxylon community and tamarix community. Furthermore, agricultural land had the most rapid expansion during the period of 1958-1968 and built-up areas showed the fastest expansion after the 2000s. (2) Two basic management modes, "local mode" formed by the local governments and "farm management mode" developed by Xinjiang Production and Construction Group, have jointly caused oasis evolution in various LUCC stages. (3) The evolution of modern oasis from the 1950s to 2004 shows the common features of driest oasis area while the period 2004-2014 is characterized by large-scale inter-basin water diversion or some new water sources. (4) The modern oasis expands at the expense of the destruction of desert vegetation, resulting in distinct variation of structure of desert plant community, which will increase the difficulty to protect desert ecosystem.

Reconstruction of the historical cropland database could provide an accurate support for the cropland researches on historical periods. Moreover, it will also act as the driving data for the global climate models. In China, there are several ways to rebuild the historical cropland on the regional scale, concerning the traditional cultivation area and the northeastern China. But there are few studies on the national scale, which is primarily due to lacking of accurate historical cropland inventory data. Therefore, it is necessary to simulate the climate and ecosystem scenarios in a wider spatial scale. In the next step, an improvement in the research on reconstruction of the historical cropland database is an efficient and quick method to meet the demand. The present paper has adopted a different approach, which divides the entire landmass of China into cropland area and livestock area according to the climate conditions and cultural diversities of the country. Several representative historical cropland databases are used as the background of our study to ensure the authenticity of the outcomes. Considering the historical land use exploitation, and the impact of the natural and anthropogenic factors on the cropland dynamics, we have developed a partition and layering-based gridded reconstruction method. This is an empirical method for spatial allocation of the historical cropland inventory data in the individual grid cells. Then we use this method to rebuild the historical cropland database maps for China during 1820 (the 25th year in the reign of Emperor Renzong Jiaqing in the Qing Dynasty) and 1936 (the 25th year of the Republic of China), at a resolution of 10 km×10 km. A precision validation is applied on the rebuilt maps by comparing them with other existing regional databases. The results show that this method could ensure the authenticity of the cropland data, with higher accuracy in the regional cropland reconstruction.

In this paper, according to the intensity of human impacts, a national hierarchical land use intensity classification system was designed based on comprehensive and high-resolution spatial datasets. According to the degree and reversibility of surface disturbance by human activities, we obtained data for four main groups of artificial land, semi-artificial land, semi-natural land and natural land, which were further divided into 22 classes based on indicators such as human population density, and cropping intensity. Land use intensity in China with a 1-km spatial resolution was obtained. The area proportions of artificial land, semi-artificial land, semi-natural land and natural land were 0.71%, 19.36%, 58.93% and 21%, respectively. Besides, significant spatial variability was demonstrated at national, regional and provincial levels. The southeastern region with denser population had more classes of land use intensity and was more intensive than the northwestern region with sparse population. With the increase of longitude, a transition in dominated land use intensity type was detected from natural and semi-natural land to semi-natural and semi-artificial land. The land use intensity in provinces of eastern China was the highest, followed by those in central and western regions. The land use intensity types showed a remarkable spatial differentiation of clustered distribution. Compared with the previous studies, our research effectively characterizes the spatial heterogeneity of land use intensity, which can help us to get a better understanding of spatial pattern of human-environment interactions and its eco-environmental impacts, which can provide scientific basis for sustainable land use policies.